Regular Patterns for Proteome-Wide Distribution of Protein Abundance across Species

doi:10.1371/journal.pone.0032423

PLoS One. 2012; 7(3): e32423.

Published online 2012 March 9. doi: 10.1371/journal.pone.0032423

PMCID: PMC3302874

Copyright Zhong et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Regular Patterns for Proteome-Wide Distribution of Protein Abundance across Species

Fan Zhong,^#^1,² Dong Yang,^#^1,³ Yunwei Hao,^#^1,³ Chengzhao Lin,^#² Ying Jiang,^1,³ Wantao Ying,^1,³ Songfeng Wu,^1,³ Yunping Zhu,^1,³ Siqi Liu,⁴ Pengyuan Yang,² Xiaohong Qian,^1,³ and Fuchu He^1,^2,³^*

¹State Key Laboratory of Proteomics, Beijing Proteome Research Center, Beijing Institute of Radiation Medicine, Beijing, China

²Institutes of Biomedical Sciences, Department of Chemistry, Fudan University, Shanghai, China

³National Engineering Research Center for Protein Drugs, Beijing, China

⁴Beijing Genomics Institute, Chinese Academy of Sciences, Beijing, China

Bin Xue, Editor

University of South Florida, United States of America

^#Contributed equally.

* E-mail: hefc/at/nic.bmi.ac.cn

Conceived and designed the experiments: FZ DY YH CL FH. Performed the experiments: FZ DY YH CL. Analyzed the data: FZ DY YH CL. Contributed reagents/materials/analysis tools: YJ WY SW YZ SL PY XQ FH. Wrote the paper: FZ DY YH CL FH.

Received November 22, 2011; Accepted January 26, 2012.

Abstract

A proteome of the bio-entity, including cell, tissue, organ, and organism, consists of proteins of diverse abundance. The principle that determines the abundance of different proteins in a proteome is of fundamental significance for an understanding of the building blocks of the bio-entity. Here, we report three regular patterns in the proteome-wide distribution of protein abundance across species such as human, mouse, fly, worm, yeast, and bacteria: in most cases, protein abundance is positively correlated with the protein's origination time or sequence conservation during evolution; it is negatively correlated with the protein's domain number and positively correlated with domain coverage in protein structure, and the correlations became stronger during the course of evolution; protein abundance can be further stratified by the function of the protein, whereby proteins that act on material conversion and transportation (mass category) are more abundant than those that act on information modulation (information category). Thus, protein abundance is intrinsically related to the protein's inherent characters of evolution, structure, and function.

Introduction

As the endpoint of the central dogma of biology, proteins are central players of most biological processes, and their abundances are regulated precisely to meet with dynamic functional requirement [1]. As the result of regulatory and evolutionary processes, a proteome may span 6 to 12 orders of magnitude among abundance distribution [2]. Exploring the principle controlling protein abundance is of great importance for the deep understanding of the building blocks of a bio-entity. Previous studies have made great progress about this open question [3]. In general, genes with higher mRNA/protein expression level tend to evolve slowly [4]–[6], have less intronic DNA [7], code for smaller proteins [8], [9], have higher biases of amino acid composition [10], [11] and codon usage [11]–[13], and tend to carry out “core functions” [11], [14]. However, these analyses are mostly based on high throughput transcript expression data, or proteome data only from single species. What' more, previous studies didn't pay attentions to the relationship between protein abundance and its character of domain composition. There is lack of systematic analysis to explore the relationship between protein abundance and protein's inherent characters at whole proteome level across species.

To this end, it is necessary to identify and quantify proteins at proteome-wide. With the fast advancement of quantitative proteomics based on mass spectrometry, protein abundance can be measured at the whole proteome level, making biological rule discovery possible. The Human Liver Proteome Project (HLPP) [15], orchestrated by the Human Proteome Organization (HUPO), is the first global survey of proteomes for a human organ, and has produced a comprehensive and reliable dataset [16]. Moreover, some quantitative proteome datasets have been generated from other species, such as M. musculus [17], [18], D. melanogaster [19], C. elegans [20], S. cerevisiae [21], and E. coli [11]. We analysed these data and uncovered three regular patterns regarding the proteome-wide distribution of protein abundance in the context of evolution, structure, and function.

Materials and Methods

1. Protein quantification

The quantification of protein abundance in the human liver was estimated using the spectral counts (SC) method. The spectral count index (SCI), an adjusted SC in accounting for the theoretic tryptic-peptide number of protein, was calculated following the reported methods [16]. To integrate data from multiple data sources, various SCI values from different experimental batches were normalized, and this normalized SCI value was termed the spectral count index normalized (SCIN) [16]. The quantitative proteomes of the mouse liver [17], D. melanogaster [19], and C. elegans [20] were also calculated according to the SCI method. Protein abundance in the mouse renal cortex [18], S. cerevisiae [21], and E. coli [11] proteome datasets were presented by their normalized spectral abundance factor (NSAF) value and copies per cell value from original papers, respectively.

2. Functional and subcellular categorization by Gene Ontology

All the gene products of seven datasets (proteomes of human liver, mouse liver, mouse kidney, D. melanogaster, C. elegans, and E. coli) were categorized into given function and subcellular classes by Gene Ontology (GO) annotations [22] according to the Generic GO Slim v1.770 standard.

To obtain some specialized functional categories such as “meiosis” and “cytokinesis”, functional and subcellular categories of yeast gene products were assigned automatically by the GO Slim Mapper system in the Saccharomyces Genome Database (SGD) according to the “Yeast GO Slim” standard.

3. Estimation of protein origin time and sequence conservation

Protein origin time was estimated by species diversity time [23]. Proteins of the seven proteome datasets were assigned orthology group IDs by OrthoMCL v2. The origin time of an orthology group was estimated by the latest diversity time that can cover all the species involved. All the 87 species in OrthoMCL were used to demarcate the orthology group origin time. The diversity time tree of all the 87 species in OrthoMCL was constructed (Fig. S1A), and then was simplified to 7 significant diversity time nodes (Fig. S1B). The protein sequences of the seven proteome datasets were extracted, and sequence conservation was determined using the bi-BLAST method with BioEdit software.

4. Functional enrichment analysis

The analysis was carried out using the web-accessible DAVID (Database for Annotation, Visualization and Integrated Discovery) tool [24]. DAVID can recognize the IPI identifier from the human liver and the mouse liver datasets, UniProt AC from the mouse kidney, the D. melanogaster and the E. coli datasets, WBGene ID (a type of ID in WormBase) from the C. elegans dataset, and the SGD (Saccharomyces Genome Database) ID from the S. cerevisiae dataset. Medium classification stringency and 12 items were chosen for enrichment calculation: COG_ONTOLOGY, SP_PIR_KEYWORDS, UP_SEQ_FEATURE, GOTERM_BP_FAT, GOTERM_MF_FAT, KEGG_PATHWAY, BIOCARTA, BBID, INTERPRO, PIR_SUPERFAMILY, SMART, and OMIM_DISEASE. To avoid high enrichment score disturbance of cellular component categories, GOTERM_CC_FAT was neglected on purpose.

5. Definition of protein domain characteristic parameters

Protein domains were defined essentially as described previously [25], using RPS-BLAST against the Pfam [26] and SMART [27] databases with the E-value An external file that holds a picture, illustration, etc.
Object name is pone.0032423.e001.jpg

An external file that holds a picture, illustration, etc.
Object name is pone.0032423.e001.jpg

after masking low-complexity regions. The domain number (DN) in one protein was calculated including the domain repeats in one protein. Domain coverage (DC) refers the percentage of the entire length of a protein occupied by domains. DC was divided by its corresponding DN to get the DC/DN value. Protein-protein interaction (PPI) domains were defined through the InterDom database [28]. PPI domain number (PPI_DN) refers the number of PPI domains in one protein, and PPI domain coverage (PPI_DC) means the percentage of the entire length of a protein occupied by PPI domains. PPI_DC/PPI_DN was calculated by dividing PPI_DC by PPI_DN.

6. Statistical analysis

All difference-tests in our analysis were carried out by using the Wilcoxon rank sum test, a non-parametric statistical hypothesis test for assessing whether two independent samples of observations have equally large values. All the correlations were defined on the non-parametric Spearman rank correlation, which assesses how well the relationship between two variables can be described using a monotonic function. We perform Wilcoxon rank sum test and spearman rank correlation using MATLAB 7.11.0. Predictive power means the probability that any two random pairs of proteins meet with the general trend. The predictive power value of each correlation was calculated based on pairs of randomly sampled proteins in each proteome dataset, and is represented by the percentage of the true predictions over all the protein pairs selected. We selected various subsets of protein pairs with a minimum rank difference percentage from 0 to 95%, and obtained the minimum and maximum predictive power values for each correlation [29].

Results

1. Evolutionary rule in protein abundance: origination and conservation versus abundance

An important evolutionary property of a protein is its origination time, which reflects its evolutionary span. Taking a proteome as a whole, we found that the abundance of proteins is positively correlated with the protein's origination time, though sometimes the correlations are not so strong (Fig. 1 and Table S1). That is, the earlier a protein is born, the higher its abundance tends to be. This tendency exists in all the six species examined, indicating that it might be a general rule of organism phylogeny. One hypothesis is that the proteins that appeared first are likely to carry out essential housekeeping roles and are, therefore, required in higher abundance, whereas proteins that evolved later tend to act in specialized functions that do not need to be in high abundance. Enrichment analysis of the functional categories in proteins supports the hypothesis that proteins of housekeeping function or regulatory functions are clustered with distinct origination time in all the six species (Table S2).

Figure 1

Proteome-wide correlation of proteins' abundance with their origin time across six species.

An interesting question in molecular evolution is why proteins evolve at different rates [4]. A negative correlation between transcript abundance and the rate of sequence evolution, in which highly expressed (transcripts level) genes evolve slowly, has been shown to hold true across the species [6], while a positive correlation between protein abundance and sequence conservation has been recently discovered from the quantitative proteomic data of C. elegans and D. Melanogaster [30]. To determine whether this phenomenon can be extended towards two evolutionary directions beyond invertebrates, we investigated mammals and single-cell organisms–species that lie in two evolutionary directions from invertebrates–and indeed found a positive correlation between abundance and sequence conservation on the proteome level among most species examined (Fig. 2). Interestingly, we found that the correlation between protein conservation degree (human vs. mouse) and protein abundance is not significant (Fig. 2, A5, B5), which was never reported before. As for comparison of human and mouse, the two species showed high conservation and the identity of most proteins is over 80%, which lead to no relationship with abundance distribution. What's more, as for E. coli, the statistic data was not powerful in case that the homologous proteins were too little in the OrthoMCL database, which may partly lead to the low correlation between conservation degree (E. coli vs. mouse) and protein abundance (Fig. 2, B1 and F4).

Figure 2

Proteome-wide correlation of proteins' abundance with their sequence conservation across six species.

2. Structural rule in protein abundance: domain versus abundance

The function of a protein is determined by its structure, which is mostly embodied in its domain architecture. We investigated the relationship between protein abundance and protein domain characteristics, including DN [31], [32] and DC [33]. DN is the number of annotated domains in the sequence of a protein, whereas DC is the percentage of the amino acid sequence that defines the domain over the whole protein sequence. As additional evidence, the number or coverage of domains mediating PPI (PPI_DN or PPI_DC) [33] are also employed in the analyses.

We found that the DN or PPI_DN of a protein is negatively correlated with its abundance, especially in higher eukaryotes (Fig. 3A and Fig. S2A). Notably, the correlations are not strong in the proteome data of S. cerevisiae and E. coli. This exception may be partly due to the fact that the proportions of multi-domain proteins in these two single-cellular organisms are much lower than other multi-cellular organisms, which is revealed by our analysis based on the data used in this study (Fig. S3A) and previous related studies [32]. We found the proportions of multi-domain proteins correlate well with the strengths of the correlations between DN and protein abundance (R

0.9252, P

0.0028, see Fig. S3B). In contrast, the DC or PPI_DC of proteins positively correlated with their abundance in the proteomes of all the six species (Fig. 3B and Fig. S2B).

Figure 3

Proteome-wide correlation between proteins' abundance and their domain characters.

DN can be regarded as the simplest parameter representing the complexity of a protein structure [31], [32], [34], and PPI_DN represents the complexity of the protein structure related to PPI. The smaller the DN or PPI_DN, the simpler the protein is. It can be inferred that proteins with simpler structures tend to be of higher abundance. Similarly, DC is an index of the structural compactness of a protein, and PPI_DC is an indicator of the structural compactness of PPI [33]. The significantly positive correlations between protein abundance and the proteins' DC or PPI_DC reveal that the more compact proteins are of higher abundance. To represent the integrated structural character, including complexity and compactness, we introduced a combined parameter, DC/DN (or PPI_DC/PPI_DN). We found that the DC/DN or PPI_DC/PPI_DN of proteins positively correlated with their abundance in the proteomes of all the six species (Fig. 3C and Fig. S2C). Thus, simple and compact proteins tend to exhibit high abundance, which does have some predictive power especially when proteins with vastly different domain characters are compared (Table S3).

Interestingly, the strength of correlation coefficient increases through the course of evolution (Fig. 3 and Fig. S2), which may be related to the increasing of biological complexity. The abundance of proteins in the higher multi-cellular organisms may need more precise regulation, to meet with the complex functional requirement. This may lead to the stronger correlation between protein abundance and its complexity or compactness.

3. Functional rule in protein abundance: function versus abundance

Cellular functions can be grouped into two categories: “mass category” and “information category”. Proteins in the mass category act on material conversion and transportation, including carbohydrate/lipid metabolism, amino acid/nucleotide metabolism, and energy metabolism; proteins in the information category act on information modulation and transduction, including response from extracellular stimulus to intracellular signal transduction, to gene regulation and expression. We found that from H. sapiens to E. coli, the abundance of proteins in the mass category is significantly higher than that of proteins in the information category (Fig. 4A–F and Tables S4, S5). For example, meiosis and cytokinesis proteins, which participate in a typical information process of S. cerevisiae that instantly decide the fate of a cell, have low abundance distributions (Fig. 4E). Many proteins in processing information have lower abundance than those in mass activities (Fig. 4G, 4H). Even in the metabolism category, which belongs to the mass category, nucleic acid and protein metabolisms, which are closer to information processing, exhibit lower abundance distributions than proteins involved in metabolisms of simpler compounds, such as lipid, amino acid, and carbohydrates (Fig. 4I, 4J).

Figure 4

Proteome-wide correlation of proteins' abundance with their functional categorization across six species.

The distinction of mass category and information category in protein abundance correlation is predicted by the evolutionary and structural rules. For example, human proteins in the mass category are highly enriched in proteins originating earlier than 4 Gya, and tend to be with lower DN, and higher DC and PPI_DC than those in the information category (Fig. S4, B1–D6). As expected from the Functional Rule in Protein Abundance, the central dogma of genetic information flow (DNA→mRNA→proteins) follows the trend of protein abundance up-regulation correlated with top/down information flow, in which proteins participating in DNA replication are of the lowest abundance, those in RNA transcription are of median abundance, and those in protein translation are of the highest abundance across species (Fig. 4K–P and Table S6). Also as predicted, along the biological information flow of “signalling→transcription→translation→metabolism”, downstream systems with higher abundance are of an earlier origination tendency than upstream ones (Fig. S4, A1–6).

Discussion

Genome sequencing has enabled us to acquire a vast amount of data, and this has accelerated the discovery of biological rules at the genome level [35], [36]. In comparison with the genome and also transcriptome, the proteome is more dynamic and diverse in composition, modification, interaction, and localization. There remain many fundamental rules or regular patterns to be discovered that may explain the diverse nature of the proteome. With the advancement of proteomics, more and more studies began to mine the rules behind the huge datasets [37].

Protein abundance is one of important phenotypic variables of proteins, and is controlled precisely. Exploration for the proteome-wide relationship between the intrinsic properties (such as evolutionary, structural or functional characters) and this phenotypic variable will benefit to discovering the essential rules related to proteins. Here, we described three correlations of protein abundance with the protein's intrinsic properties of evolution, structure, and function, observed to be consistent from bacteria to yeast, worm, fly, mouse, and human: (1) There is a positive correlation with both the protein's origination time and sequence conservation during phylogeny, confirming the conclusions of previous studies at a wider range of species [4]–[6]; (2) We found protein abundance negatively correlated with the protein's domain number, and positively correlated with protein's domain coverage, which indicates that proteins with simple/compact structures tend to be high abundant. The finding that proteins with more domains tend to be with lower abundance may be due to the selection to reduce the risk of mis-translation [6] of the multi-domain proteins. Another phenomenon that higher abundant proteins tend to have higher domain coverage may result from the selection to reduce the length of non-functional regions within the abundant proteins to minimize transcriptional and translational costs, just as the similar selection on the length of intronic DNA [7]. (3) The abundances of proteins involved in the mass category tend to be higher than those in the information category, which also is similar to previous studies' conclusions obtained from the data of bacteria [11] or human cell line [14]. Our work firstly confirmed that this finding is presented across species, and some new findings emerged based on our new analyses. Intriguingly, the third pattern can be inferred from the first two patterns, highlighting the importance of the structure and function relationship.

Taken together, protein abundance distribution across the whole proteome displays prominent regularities, even though there are large discrepancies in protein composition and abundance among various proteomes of diverse bio-entities. Such regularities seem to be maintained in different organs as analysis of the mouse kidney dataset yielded identical patterns with the analysis of the mouse liver datasets (Fig. S5). The discovery of those regularities strongly demonstrated that the quantitative and comprehensive proteomic datasets could provide a rich ground for the exploration and discovery of the fundamental rules in nature on the proteome scale.

Supporting Information

Figure S1

Diversity time of all 87 species in OrthoMCL v2. (A) Detailed diversity time tree that was constructed according to the reference (Hedges 2002). (B) Simplified diversity time tree with 7 significant time nodes to keep proper time scale distances (0.58 or 0.37, 0.26, 0.39, 1.77 and 1.40 Gya respectively) for analysis. The 6 species analyzed were in red text.

(PNG)

Click here for additional data file.^{(312K, png)}

Figure S2

Proteome-wide correlation between protein abundance and its protein-protein interacting (PPI) domain parameters. Three Parameters including PPI domain number (PPI_DN) (A), PPI domain coverage (PPI_DC) (B) and PPI_DC/PPI_DN (C) were employed for the analyses. R represents Spearman rank correlation coefficient and P represents the P-value of the spearman rank correlation analysis. Medians are indicated as black dots (A), or crosses (B, C), and whiskers encompass the range from 25% to 75% of values.

(TIF)

Click here for additional data file.^{(189K, tif)}

Figure S3

The proportion of multi-domain proteins in each dataset (A) and its relationship with the correlation coefficient values between domain number and protein abundance (B). E. coli (Ec), S. cerevisiae (Sc), C. elegans (Ce), D. melanogaster (Dm), M. musculus cortex of kidney (MmC), M. musculus liver (MmL),and H. sapiens (Hs). R represents the pearson correlation coefficient, and P represent the P-value.

(PNG)

Click here for additional data file.^{(30K, png)}

Figure S4

Comparison of mass categorial proteins with information categorial ones on distributions of origin time and domain character among six species. Origin time distributions of proteins in H. sapiens (a1), M. musculus (a2), D. melanogaster (a3), C. elegans (a4), S. cerevisiae (a5) and E. coli (a6). Domain character distributions of proteins in the same species: DN (b1–6), DC (c1–6) and PPI_DC (d1–6).

(PNG)

Click here for additional data file.^{(922K, png)}

Figure S5

Consistency of the three rules across tissues. When replacing mouse kidney data by mouse liver data in analyses, all rules were maintained. These analyses include: correlations of protein abundance with origin time (A), sequence conservation (B) domain characters (C), functional categories (D) and biogenesis machines of three bio-molecules (E).

(TIF)

Click here for additional data file.^{(750K, tif)}

Table S1

Rank sum test p-values between origin time categorized proteins' abundance datasets.

(DOC)

Click here for additional data file.^{(78K, doc)}

Table S2

The functional enrichment results of different origin time protein categories in 6 species by DAVID.

(DOC)

Click here for additional data file.^{(44K, doc)}

Table S3

Predictive power of each domain character parameter correlating with protein abundance.

(DOC)

Click here for additional data file.^{(44K, doc)}

Table S4

Rank sum test p-values between various functional categorized proteins' abundance datasets across six species.

(DOC)

Click here for additional data file.^{(105K, doc)}

Table S5

Rank sum test p-values between various functional categorized proteins' abundance datasets in H. sapiens and S. cerevisiae.

(DOC)

Click here for additional data file.^{(57K, doc)}

Table S6

Rank sum test p-values between proteins' abundance datasets of three bio-molecules biogenesis machines across six species.

(DOC)

Click here for additional data file.^{(36K, doc)}

Acknowledgments

We thank Profs. Jun Qin and Chung-I Wu for scientific and linguistic advice, Mr. Hailin Tang for programming, Ms. Xue Gao, Mr. Junjie Zheng, Ms. Xueli Zhang for logistic support, and all the members of CNHLPP Consortium and HLPP Consortium.

Footnotes

Competing Interests: The authors have declared that no competing interests exist.

Funding: This work was partially supported by Chinese State Key Projects for Basic Research (“973 Program”) (Nos. 2010CB912704, 2011CB910601 and 2011CB910700); National Natural Science Foundation of China (81123001, 30900257, 30972909, 31000379, 81000192, 81001470, 81170399 and 81010064); International Scientific Collaboration Program (2009DFB33070, 2010DFA31260 and 2011DFB30370); State Key Laboratory of Proteomics (SKLP-O201101); Chinese State Key Project Specialized for Infectious Diseases (2012ZX10002012-006); Shanghai Health Bureau Project (2010Y005) and the Natural Science Foundation of Shanghai (No. 09ZR1404100). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1. Cox J, Mann M. Is proteomics the new genomics? Cell. 2007;130:395–398. [PubMed]

2. Eriksson J, Fenyö D. Improving the success rate of proteome analysis by modeling protein-abundance distributions and experimental designs. Nat Biotechnol. 2007;25(6):651–655. [PubMed]

3. Yang D, Jiang Y, He F. An integrated view of the correlations between genomic and phenomic variables. J Genet Genomics. 2009;36:645–651. [PubMed]

4. Drummond DA, Bloom JD, Adami C, Wilke CO, Arnold FH. Why highly expressed proteins evolve slowly? Proc Natl Acad Sci USA. 2005;102:14338–14343. [PubMed]

5. Tuller T, Kupiec M, Ruppin E. Determinants of protein abundance and translation efficiency in S. cerevisiae. PLoS Comput Biol. 2007;3:e248. [PMC free article] [PubMed]

6. Drummond DA, Wilke CO. Mistranslation-induced protein misfolding as a dominant constraint on coding-sequence evolution. Cell. 2008;134:341–352. [PMC free article] [PubMed]

7. Castillo-Davis CI, Mekhedov SL, Hartl DL, Koonin EV, Kondrashov FA. Selection for short introns in highly expressed genes. Nat Genet. 2002;31:415–418. [PubMed]

8. Urrutia AO, Hurst LD. The signature of selection mediated by expression on human genes. Genome Res. 2003;3:2260–2264. [PubMed]

9. Warringer J, Blomberg A. Evolutionary constraints on yeast protein size. BMC Evol Biol. 2006;6:61. [PMC free article] [PubMed]

10. Greenbaum D, Jansen R, Gerstein M. Analysis of mRNA expression and protein abundance data: An approach for the comparison of the enrichment of features in the cellular population of proteins and transcripts. Bioinformatics. 2002;18:585–596. [PubMed]

11. Ishihama Y, Schmidt T, Rappsilber J, Mann M, Hartl FU, et al. Protein abundance profiling of the Escherichia coli cytosol. BMC Genomics. 2008;9:102. [PMC free article] [PubMed]

12. Ingvarsson PK. Gene expression and protein length influence codon usage and rates of sequence evolution in Populus tremula. Mol Biol Evol. 2007;24:836–844. [PubMed]

13. Urrutia AO, Hurst LD. Codon bias covaries with expression breadth and the rate of synonymous evolution in humans, but this is not evidence for selection. Genetics. 2001;159:1191–1199. [PubMed]

14. Beck M, Schmidt A, Malmstroem J, Claassen M, Ori A, et al. The quantitative proteome of a human cell line. Mol Syst Biol. 2011;7:549. [PMC free article] [PubMed]

15. He F. Human liver proteome project: plan progress and perspectives. Mol Cell Proteomics. 2005;4:1841–1848. [PubMed]

16. Jiang Y, Ying W, Wu S, Chen M, Zhong F, et al. First insight into the human liver proteome from PROTEOME(SKY)-LIVER(Hu) 1.0 a publicly available database. J Proteome Res. 2010;9:79–94. [PubMed]

17. Lai KK, Kolippakkam D, Beretta L. Comprehensive and quantitative proteome profiling of the mouse liver and plasma. Hepatology. 2008;47:1043–1051. [PubMed]

18. Zhao Y, Denner L, Haidacher SJ, LeJeune WS, Tilton RG. Comprehensive analysis of the mouse renal cortex using two-dimensional HPLC-tandem mass spectrometry. Proteome Sci. 2008;6:15. [PMC free article] [PubMed]

19. Brunner E, Ahrens CH, Mohanty S, Baetschmann H, Loevenich S, et al. A high-quality catalog of the Drosophila melanogaster proteome. Nat Biotechnol. 2007;25:576–583. [PubMed]

20. Merrihew GE, Davis C, Ewing B, Williams G, Kall L, et al. Use of shotgun proteomics for the identification confirmation and correction of C elegans gene annotations. Genome Res. 2008;18:1660–1669. [PubMed]

21. Ghaemmaghami S, Huh WK, Bower K, Howson RW, Belle A, et al. Global analysis of protein expression in yeast. Nature. 2003;425:737–741. [PubMed]

22. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet. 2000;25:25–29. [PMC free article] [PubMed]

23. Hedges SB. The origin and evolution of model organisms. Nat Rev Genet. 2002;3:838–849. [PubMed]

24. Huang da W, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009;4:44–57. [PubMed]

25. Basu MK, Carmel L, Rogozin IB, Koonin EV. Evolution of protein domain promiscuity in eukaryotes. Genome Res. 2008;18:449–461. [PubMed]

26. Finn RD, Tate J, Mistry J, Coggill PC, Sammut SJ, et al. The Pfam protein families database. Nucleic Acids Res. 2008;36(Database issue):D281–288. [PMC free article] [PubMed]

27. Letunic I, Copley RR, Pils B, Pinkert S, Schultz J, et al. SMART 5: domains in the context of genomes and networks. Nucleic Acids Res. 2006;34(Database issue):D257–260. [PMC free article] [PubMed]

28. Ng SK, Zhang Z, Tan SH, Lin K. InterDom: a database of putative interacting protein domains for validating predicted protein interactions and complexes. Nucleic Acids Res. 2003;31:251–254. [PMC free article] [PubMed]

29. Wang Z, Zhang J. Why is the correlation between gene importance and gene evolutionary rate so weak? PLoS Genet. 2009;5:e1000329. [PMC free article] [PubMed]

30. Schrimpf SP, Weiss M, Reiter L, Ahrens CH, Jovanovic M, et al. Comparative functional analysis of the Caenorhabditis elegans and Drosophila melanogaster proteomes. PLoS Biol. 2009;7:e48. [PMC free article] [PubMed]

31. Koonin EV, Aravind L, Kondrashov AS. The impact of comparative genomics on our understanding of evolution. Cell. 2000;101:573–576. [PubMed]

32. Koonin EV, Wolf YI, Karev GP. The structure of the protein universe and genome evolution. Nature. 2002;420:218–223. [PubMed]

33. Xia K, Fu Z, Hou L, Han JD. Impacts of protein-protein interaction domains on organism and network complexity. Genome Res. 2008;18:1500–1508. [PubMed]

34. Oakley TH, Rivera AS. Genomics and the evolutionary origins of nervous system complexity. Curr Opin Genet Dev. 2008;18:479–492. [PubMed]

35. Lander ES, Linton LM, Birren B, Nusbaum C, Zody MC, et al. International Human Genome Sequencing Consortium. Initial sequencing and analysis of the human genome. Nature. 2001;409:860–921. [PubMed]

36. Altshuler D, Daly M J, Lander ES. Genetic mapping in human disease. Science. 2008;322:881–888. [PMC free article] [PubMed]

37. Saeed R, Deane CM. Protein protein interactions, evolutionary rate, abundance and age. BMC Bioinformatics. 2006;7:128. [PMC free article] [PubMed]

Articles from PLoS ONE are provided here courtesy of
Public Library of Science